arXiv subjects

Jeff Brown

Publications and source records attributed to Jeff Brown.

4 recordsLinked to original sources

arXiv2026

ConnectomeBench2: A Unified Benchmark for Automated Connectomic Proofreading

Proofreading--correcting segmentation errors in 3D brain reconstructions--is the rate-limiting step in synapse-resolution connectomics. We release ConnectomeBench2, a unified multi-species dataset of over 716,485 expert-labeled proofreading decisions with >4,500,000 associated images spanning four major open connectomes (mouse, human, zebrafish, fly), spanning both split and merge error correction. Trained on this dataset, a single Vision Transformer with shared encoders for mesh geometry and electron microscopy reaches human-level accuracy across species for split error correction and merge error identification, with performance scaling with data size and modality. Beyond accuracy, we show that the model is well-calibrated within distribution, that measures of distribution distance predict where calibration and accuracy will degrade on unseen data, and that connectomics-specific pretraining and active learning-based sample selection show potential to substantially reduce the labeling effort needed to extend to new species and brain regions. The benchmark provides the infrastructure to train and evaluate increasingly capable vision models for connectomic proofreading. Data and code availability. The ConnectomeBench2 dataset is released on Hugging Face at https://huggingface.co/datasets/jeffbbrown2/ConnectomeBench2. The accompanying codebase is available on GitHub at https://github.com/timfarkas/ConnectomeBench2.

cs.CV↗

arXiv2025

ConnectomeBench: Can LLMs Proofread the Connectome?

Connectomics - the mapping of neural connections in an organism's brain - currently requires extraordinary human effort to proofread the data collected from imaging and machine-learning assisted segmentation. With the growing excitement around using AI agents to automate important scientific tasks, we explore whether current AI systems can perform multiple tasks necessary for data proofreading. We introduce ConnectomeBench, a multimodal benchmark evaluating large language model (LLM) capabilities in three critical proofreading tasks: segment type identification, split error correction, and merge error detection. Using expert annotated data from two large open-source datasets - a cubic millimeter of mouse visual cortex and the complete Drosophila brain - we evaluate proprietary multimodal LLMs including Claude 3.7/4 Sonnet, o4-mini, GPT-4.1, GPT-4o, as well as open source models like InternVL-3 and NVLM. Our results demonstrate that current models achieve surprisingly high performance in segment identification (52-82% balanced accuracy vs. 20-25% chance) and binary/multiple choice split error correction (75-85% accuracy vs. 50% chance) while generally struggling on merge error identification tasks. Overall, while the best models still lag behind expert performance, they demonstrate promising capabilities that could eventually enable them to augment and potentially replace human proofreading in connectomics. Project page: https://github.com/jffbrwn2/ConnectomeBench and Dataset https://huggingface.co/datasets/jeffbbrown2/ConnectomeBench/tree/main

q-bio.NC↗

arXiv2020

Transferrable AlGaN/GaN HEMTs to Arbitrary Substrates via a Two-dimensional Boron Nitride Release Layer

Mechanical transfer of high performing thin film devices onto arbitrary substrates represents an exciting opportunity to improve device performance, explore non-traditional manufacturing approaches, and paves the way for soft, conformal, and flexible electronics. Using a two-dimensional (2D) boron nitride (BN) release layer, we demonstrate the transfer of AlGaN/GaN high-electron mobility transistors (HEMTs) to arbitrary substrates through both direct van der Waals (vdW) bonding and with a polymer adhesive interlayer. No device degradation was observed due to the transfer process, and a significant reduction in device temperature (327 {\deg}C to 132 {\deg}C at 600 mW) was observed when directly bonded to a silicon carbide (SiC) wafer relative to the starting wafer. With the use of a benzocyclobutene (BCB) adhesion interlayer, devices were easily transferred and characterized on Kapton and ceramic films, representing an exciting opportunity for integration onto arbitrary substrates. Upon reduction of this polymer adhesive layer thickness, the AlGaN/GaN HEMTs transferred onto a BCB/SiC substrate resulted in comparable peak temperatures during operation at powers as high as 600 mW to the as-grown wafer, revealing that by optimizing interlayer characteristics such as thickness and thermal conductivity, transferrable devices on polymer layers can still improve performance outputs.

physics.app-ph↗

arXiv2015

A Mirror Theorem for T-Equivariant Blowups

Let E be a toric fibration arising from symplectic reduction of a direct sum of line bundles over (almost-) K\"ahler base B. Then each torus-fixed point of the toric manifold fiber defines a section of the fibration. Let L_a be convex line bundles over B, A_a smooth divisors of B arising as the zero loci of generic sections of L_a, and \a:B\to E a particular fixed-point section of E. Further assume the \{A_a\} to be mutually disjoint. We compute genus-0 Gromov--Witten invariants of the blowup of E along \a(\coprod_a A_a) in terms of genus-0 Gromov--Witten invariants of B and of \{A_a\}, the matrix used for the symplectic reduction description of the fiber of the toric fibration E\to B, and the restriction maps i_{A_a}^*:H^*(B)\to H^*(A_a).

math.AG↗